Application

Categories based Portfolio

Categories based Portfolio is an AI chat interface for building Markowitz optimized crypto portfolios by category - ask in plain language, get back an allocation with risk metrics and correlation data.

Last updated:

Categories based Portfolio is an AI-powered portfolio construction tool available at /portfolio. You describe what kind of portfolio you want in plain language, and the AI returns a Markowitz-optimized allocation across crypto categories with risk metrics.

It uses the same optimization framework as Portfolio MCP - which covers the theory in detail. This page explains how to use the chat interface.

How to use it

Type a question or instruction into the chat input and press Send or hit Enter. The AI responds with an optimized portfolio breakdown rendered as structured text.

Example prompts to get started:

  • Create a long only portfolio with 5 crypto categories
  • Show me the correlation matrix for top 5 categories
  • Create a long only portfolio with 5 interest bearing categories

You can continue the conversation - ask follow-up questions, adjust parameters, or request a different number of categories. Use Clear to reset the chat and start fresh.

What the AI can return

Depending on your prompt, the AI can produce:

  • Portfolio allocation - a breakdown across crypto categories with weights optimized for maximum Sharpe Ratio
  • Performance metrics - expected annual return, volatility, Sharpe Ratio, and Sortino Ratio for the proposed portfolio
  • Correlation matrix - pairwise correlation between selected categories, showing which assets move together and which provide genuine diversification
  • Diversification analysis - whether the spread across categories represents true diversification or correlated exposure under different names

Categories as sectors

The tool groups the crypto market into categories - such as DeFi, Layer 2, Payments, Stablecoins, and others - and treats each category as a sector, similar to how institutional equity portfolios allocate across industry sectors.

This approach reduces single-token concentration risk and allows the optimizer to work at the level of market segments rather than individual tokens. See Portfolio MCP - CoinGecko Categories as Sectors for more detail.

Interest-bearing portfolios

Including "interest bearing" in your prompt (e.g. 5 interest bearing categories) focuses the optimizer on yield-generating segments of the market - DeFi lending, staking, and similar categories - rather than pure price-appreciation assets. This is useful when your goal is income from your crypto holdings, not just capital growth.

Rebalancing

The optimizer produces a snapshot allocation based on historical return and correlation data. As markets move, actual weights drift from the optimized target. Rebalance periodically - monthly is a practical cadence for most users - by running a fresh prompt and adjusting holdings to match the new output.

Categories based Portfolio vs Broad Portfolio

Categories based Portfolio Broad Portfolio
Interface AI chat Form (enter token count)
Optimization unit Crypto categories (sectors) Individual tokens
Output Weights, metrics, correlation matrix - as chat responses Token weights table, Sharpe/Volatility charts, simulation
Best for Exploring allocation strategies, interest-bearing portfolios, sector-level diversification Getting a direct token-level portfolio for a specific number of assets

Further info